Empowerment-based self-defense (ESD) programs have proven effective in preventing sexual violence (SV) among girls in diverse settings, yet their effectiveness in South Africa remains unexplored. In this hybrid type 1 cluster-randomized controlled trial, we assessed the impact and implementation of the COVID-adapted 'No Means No' intervention, an ESD program to prevent SV among girls aged 10-19 in Gqeberha, South Africa. Fifteen schools were randomly assigned to one of three trial arms: an arm receiving intervention delivered to girls-only, one receiving interventions delivered to girls and boys seperately, or to serve as controls in a 1:1:1 ratio. Surveys were conducted at baseline, 3, 6, and 12 months. The primary outcome was past-year SV exposure among girls, including harassment or rape. Generlized Estimating Equation models compared the two intervention arms to the control, and each intervention arm to the control. Implementation data consisted of intervention attendence logs, quarterly implementation reports, and in-depth interviews with school stakeholders. We enrolled 1,540 from 14 schools, of whom 1,250 provided primary outcome data. The average age was 13 years and 83% identified as Black. At baseline, 35% of girls reported past-year SV, and 33% did during follow-up. The intervention did not significantly reduce SV compared to the control (adjRR 1.24, 95% CI 0.96, 1.69
p = 0.08). Exploratory analyses examined the effects on secondary outcomes (rape, offline, and online sexual harassment), and on intermediate outcomes (e.g., knowledge and attitudes) with only knowledge of self-defense differing between intervention and control (coefficient 0.42, 95% CI 0.16, 0.68
p = 0.007). Implementation data revealed barriers to implementing in schools post-COVID, including limitations on alloted time and lack of whole-school buy-in. Improved school awareness, willingness to disclose violence, and reductions in bullying were described as impacts of the intervention by stakeholders. While the lack of reduction in SV may point towards areas for improvement, numerous contextual and implementation factors may also have influenced results. Future trials should utilize implementation science methods to improve delivery and rigorously evaluate ESD interventions' impact on disclosure, acknowledging the complexity of assessing their effects on various facets of SV. The trial was registered on clinicaltrials.gov (NCT05295342).